Category Archives: Nutanix

Acropolis: Scalability

Posted on by
Reply

One of the major focuses for Nutanix both for our Distributed Storage Fabric (part of the Nutanix Xtreme Computing Platform or XCP) has been scalability with consistent performance.

Predictable scalability is critical to any distributed platform as it predictable scalability for the management layer.

This is one of the many strengths of the Acropolis management layer.

All components which are required to Configure, Manage, Monitor, Scale and Automate are fully distributed across all nodes within the cluster.

As a result, there is no single point of failure with the Nutanix/Acropolis management layer.

Lets take a look at a typical four node cluster:

Below we see four Controller VMs (CVMs) which service one node each. In the cluster we have an Acropolis Master along with multiple Acropolis Slave instances.

Acropolis4nodecluster1

In the event the Acropolis Master becomes unavailable for any reason, an election will take place and one of the Acropolis Slaves will be promoted to Master.

This can be achieved because Acropolis data is stored in a fully distributed Cassandra database which is protected by the Distributed Storage Fabric.

When an additional Nutanix node is added to the cluster, an Acropolis Slave is also added which allows the workload of managing the cluster to be distributed, therefore ensuring management never becomes a point of contention.Acropolis5NodeCluster

Things like performance monitoring, stats collection, Virtual Machine console proxy connections are just a few of the management tasks which are serviced by Master and Slave instances.

Another advantage of Acropolis is that the management layer never needs to be sized or scaled manually. There is no vApp/s , Database Server/s, Windows instances to deploy, install, configure, manage or license, therefore reducing cost and simplifying management of the environment.

Summary:

Acropolis Management is automatically scaled as nodes are added to the cluster, therefore increasing consistency , resiliency, performance and eliminating potential for architectural (sizing) errors which may impact manageability.

Note: For non-Acropolis deployments, PRISM is also scaled in the same manner as described above, however the scalability of Hypervisor management layers such as vCenter or SCVMM will need to be considered separately when not using Acropolis.

What’s .NEXT? – Erasure Coding!

Posted on by
9

Up to now, Nutanix has used a concept known as “Replication Factor” or “RF” to provide storage layer data protection as opposed to older RAID technologies.

RF allows customers to configure either 2 or 3 copies of data depending on how critical the data is.

When using RF2, the usable capacity of RAW is 50% (RAW divide 2).

When using RF3, the usable capacity of RAW is 33% (RAW divide 3).

While these sound like large overheads, but in reality, they are comparable to traditional SAN/NAS deployments as explain in the two part post – Calculating Actual Usable capacity? It’s not as simple as you might think!

But enough on existing features, lets talk about an exciting new feature, Erasure coding!

Erasure coding (EC) is a technology which significantly increases the usable capacity in a Nutanix environment compared to RF2.

The overhead for EC depends on the cluster size but for clusters of 6 nodes or more it results in only a 1.25x overhead compared to 2x for RF2 and 3x for RF3.

For clusters of 3 to 4 nodes, the overhead is 1.5 and for clusters of 5 nodes 1.33.

The following shows a comparison between RF2 and EC for various cluster sizes.ErasureCodingAs you can see, the usable capacity is significantly increased when using Erasure Coding.

Now for more good news, in-line with Nutanix Uncompromisingly Simple philosophy, Erasure Coding can be enabled on existing Nutanix containers on the fly without downtime or the requirement to migrate data.

This means with a simple One-click upgrade to NOS 4.5, customers can get up to a 60% increase in usable capacity in addition to existing data reduction savings. e.g.: Compression.

So there you have it, more usable capacity for Nutanix customers with a non disruptive one click software upgrade…. (your welcome!).

For customers considering Nutanix, your cost per GB just dropped significantly!

Want more? Check out how to scale storage capacity separately from compute with Nutanix!

Related Articles:

1. Nutanix Erasure Coding (EC-X) Deep Dive

What’s .NEXT? – Scale Storage separately to Compute on Nutanix!

Posted on by
2

Since I joined Nutanix, I have heard from customers that they want to scale storage (capacity) separate to compute as they have done in traditional SAN/NAS environments.

I wrote an article a while ago about Scaling problems with traditional shared storage which discusses why scaling storage capacity separately can be problematic. As such I still believe scaling capacity separately is more of a perceived advantage than a real one in most cases, especially with traditional SAN/NAS.

However here at Nutanix we have locked ourselves away and brainstormed how we can scale capacity without degrading performance and without loosing the benefits of a Nutanix Hyper-Converged platform such as Data Locality and linear scalability.

At the same time, we wanted to ensure doing so didn’t add any unnecessary cost.

Introducing the NX-6035c , a new “Storage only” node!

What is it?

The NX-6035c is a 2 node per 2 RU block, which has 2 single socket servers with 1 SSD and 5 x 3.5″ SATA HDDs and 2 x 10GB NICs for network connectivity.

How does it work?

As with all Nutanix nodes, the NX-6035c runs the Nutanix Controller VM (CVM) which presents the local storage to the Nutanix Distributed File System (NDFS).

The main difference between the NX-6035c and other Nutanix nodes is that it is not a member of the hypervisor cluster and as a result does not run virtual machines, but it is a fully functional member of the NDFS cluster.

The below diagram shows a 3 node vSphere or Hyper-V cluster with storage presented by a 5 node NDFS cluster using 3 x NX-8150s as Compute+Storage and 2 x NX-6035C nodes as Storage only.

6035cinndfscluster

Because the NX-6035c does not run VMs, it only receives data via Write I/O replication from Resliency Factor 2 or 3 and Disk Balancing.

This means for every NX-6035c in an NDFS cluster, the Write performance for the cluster increases because of the additional CVM. This is how Nutanix ensures we avoid the traditional capacity scaling issues of SAN/NAS.

Rule of thumb: Don’t scale capacity without scaling storage controllers!

The CVM running on the NX-6035c also provides data reduction capabilities just like other Nutanix nodes, so data reduction can occur with even lower impact on Virtual Machine I/O.

What about Hypervisor licensing?

The NX-6035c runs the CVM on a Nutanix optimized version of KVM which does not require any hypervisor licensing.

For customers using vSphere or Hyper-V, the NX-6035c provides storage performance and capacity to the NDFS cluster which serves the hypervisor.

This results is more storage capacity and performance with no additional hypervisor costs.

Want more? Check out how Nutanix is increasing usable capacity with Erasure Coding!